The Agent Skills Directory

[SAFE]: The skill instructions and associated language references focus on passive static analysis and documentation of existing source code. The methodology emphasizes literal code representation ('Code is Truth') and requires every diagram node to be mapped to a specific source line, which significantly reduces the risk of malicious hallucination or instruction override.
[INDIRECT_PROMPT_INJECTION]: The skill processes external source code and manifest data, creating an ingestion surface for indirect prompt injection.
Ingestion points: Reads targets from docs/reverse/{analysis}/manifest.json (Step 1) and ingests source code via Read and symbol lookup tools (Step 2).
Boundary markers: None explicitly defined for the source code ingestion, though the requirement for line-number traceability acts as a functional constraint.
Capability inventory: The skill uses Write to generate markdown reports and update the project manifest (Step 5, Step 6).
Sanitization: The skill implements specific Mermaid syntax rules in Step 3 to escape special characters and handle generic/array syntax, which serves to sanitize the data before it is rendered in the final documentation.

re-visualize-logic